Cross-sectional associations between air pollution ...

Downloaded from thorax.bmj.com on August 15, 2014 - Published by group.bmj.com

Thorax Online First, published on August 11, 2014 as 10.1136/thoraxjnl-2013-204352 Environmental exposure

ORIGINAL ARTICLE

Cross-sectional associations between air pollution and chronic bronchitis: an ESCAPE meta-analysis across five cohorts Yutong Cai,1 Tamara Schikowski,2,3,4 Martin Adam,2,3 Anna Buschka,4 Anne-Elie Carsin,5 Benedicte Jacquemin,5,6,7 Alessandro Marcon,8 Margaux Sanchez,6,7 Andrea Vierkötter,4 Zaina Al-Kanaani,1 Rob Beelen,9 Matthias Birk,10 Bert Brunekreef,9 Marta Cirach,5 Françoise Clavel-Chapelon,7,11 Christophe Declercq,12,† Kees de Hoogh,1,2,3 Audrey de Nazelle,5,13 Regina E Ducret-Stich,2,3 Virginia Valeria Ferretti,14 Bertil Forsberg,15 Margaret W Gerbase,16 Rebecca Hardy,17 Joachim Heinrich,10 Gerard Hoek,9 Debbie Jarvis,1,18 Dirk Keidel,2,3 Diana Kuh,17 Mark J Nieuwenhuijsen,5 Martina S Ragettli,2,3 Andrea Ranzi,19 Thierry Rochat,16 Christian Schindler,2,3 Dorothea Sugiri,4 Sofia Temam,6,7 Ming-Yi Tsai,2,3 Raphaëlle Varraso,6,7 Francine Kauffmann,6,7 Ursula Krämer,4 Jordi Sunyer,5 Nino Künzli,2,3 Nicole Probst-Hensch,2,3 Anna L Hansell1,20 ▸ Additional material is published online only. To view please visit the journal online (http://dx.doi.org/10.1136/ thoraxjnl-2013-204352). For numbered affiliations see end of article. Correspondence to Dr Anna Hansell, MRC-PHE Centre for Environment and Health, School of Public Health, Imperial College London, St Mary’s Campus, Norfolk Place, London W2 1PG, UK; [email protected] YC, TS and MA contributed equally to this work. TS, FK UK, JS, NK, NP-H and ALH: Steering Committee ESCAPE Work Package 4 Respiratory Health in Adults. †Deceased. Received 14 August 2013 Revised 11 July 2014 Accepted 17 July 2014

To cite: Cai Y, Schikowski T, Adam M, et al. Thorax Published Online First: [ please include Day Month Year] doi:10.1136/thoraxjnl-2013204352

ABSTRACT Background This study aimed to assess associations of outdoor air pollution on prevalence of chronic bronchitis symptoms in adults in five cohort studies (Asthma-E3N, ECRHS, NSHD, SALIA, SAPALDIA) participating in the European Study of Cohorts for Air Pollution Effects (ESCAPE) project. Methods Annual average particulate matter (PM10, PM2.5, PMabsorbance, PMcoarse), NO2, nitrogen oxides (NOx) and road traffic measures modelled from ESCAPE measurement campaigns 2008–2011 were assigned to home address at most recent assessments (1998–2011). Symptoms examined were chronic bronchitis (cough and phlegm for ≥3 months of the year for ≥2 years), chronic cough (with/without phlegm) and chronic phlegm (with/ without cough). Cohort-specific cross-sectional multivariable logistic regression analyses were conducted using common confounder sets (age, sex, smoking, interview season, education), followed by meta-analysis. Results 15 279 and 10 537 participants respectively were included in the main NO2 and PM analyses at assessments in 1998–2011. Overall, there were no statistically significant associations with any air pollutant or traffic exposure. Sensitivity analyses including in asthmatics only, females only or using back-extrapolated NO2 and PM10 for assessments in 1985–2002 (ECRHS, NSHD, SALIA, SAPALDIA) did not alter conclusions. In never-smokers, all associations were positive, but reached statistical significance only for chronic phlegm with PMcoarse OR 1.31 (1.05 to 1.64) per 5 mg/m3 increase and PM10 with similar effect size. Sensitivity analyses of older cohorts showed increased risk of chronic cough with PM2.5abs (black carbon) exposures. Conclusions Results do not show consistent associations between chronic bronchitis symptoms and current traffic-related air pollution in adult European populations.

Key messages What is the key question? ▸ Is long-term exposure to traffic or ambient air pollution associated with prevalence of cough and phlegm in adult European populations?

What is the bottom line?

▸ Current long-term average air pollution levels were not associated with symptoms of chronic bronchitis, cough or phlegm in European adults of all ages living in nine European countries, but there were small increases in reported phlegm in never-smokers associated with coarse particulate matter.

Why read on? ▸ This is one of the largest such studies in adults involving >10 000 individuals in five European cohorts using harmonised exposure and outcome measurements; while most results were null, there was some heterogeneity across findings for cohort assessments at different time points, particularly for black carbon and NO2.

INTRODUCTION Chronic cough and phlegm production are common respiratory symptoms. In the past, these were often considered together as the clinical phenotype of chronic bronchitis,1 but more recently phlegm2 and cough3 have been considered separately and may have differing mechanisms—for example, cough may result from central reflex sensitivity4 as well as irritation and inflammation. A previous study of young adults found wide geographic variability in

Cai Y, et al. Thorax 2014;0:1–10. doi:10.1136/thoraxjnl-2013-204352 1 Copyright Article author (or their employer) 2014. Produced by BMJ Publishing Group Ltd (& BTS) under licence.


Environmental exposure chronic bronchitis prevalence (0.7–9.7%) across Europe, but only 30% of the variability could be explained by differences in smoking habits.5 This suggests other potentially modifiable factors—such as air pollution—may be important. There is good evidence that air pollution triggers exacerbations in patients with COPD, and suggestive evidence of chronic effects of air pollution on the prevalence and incidence of COPD in adults.6 Concurrent asthma may give rise to cough and phlegm. Traffic-related air pollution has also been related to onset of childhood asthma, but findings in adults are less clear.7 Pathophysiological studies have found associations between long-term exposure to ambient particulate matter (PM) and chronic mucosal inflammation in the lung,8 resulting in excessive mucus secretion, coughing and phlegm production.9 Previous epidemiological studies examining associations between objectively measured air pollution and chronic bronchitis symptoms in adults10–27 are difficult to compare. For example, some studies have used surrogate measures for air pollution (eg, distance from the main road,13 20 21 27 traffic intensity17), others used air pollution data from local monitoring networks10–13 16 22 23 25 26 or model-derived exposures estimated at home address.15 17 19 24 Some13 15 17 but not all14 18 19 23 studies reported increased risks in the general population, whereas studies in specific populations reported associations only in never-smokers10–12 16 26 or females.20 21 25 The present study investigates cross-sectional associations between ambient air pollution estimated at home address and prevalence of chronic bronchitis symptoms in five European cohort studies participating in European Study of Cohorts for Air Pollution Effects (ESCAPE) project. Taking advantage of individual information and repeated assessments, we gave special attention to the time period of exposure (contemporary, historic 2000s and historic 1990s exposures) in repeated crosssectional analyses and conducted extensive sensitivity analyses.

METHODS Study populations Analyses were based on subpopulations from the European Community Respiratory Health Survey (ECRHS); National Survey of Health and Development (NSHD) from the UK; the Study on the influence of Air pollution on Lung function, Inflammation and Aging (SALIA) from Ruhr area in Germany; the Swiss cohort study on Air Pollution And Lung and heart Diseases in Adults (SAPALDIA); and the French Asthma-E3N study, an asthma case–control study nested in the ‘Etude Épidémiologique de Femmes de la Mutuelle Générale de l’Education Nationale (E3N)’ cohort who were living in geographic areas covered by ESCAPE exposure models (‘ESCAPE areas’). A brief description of each cohort is available in online supplements-1. Those included in the analyses had valid chronic bronchitis data and information on sex, age, smoking status, season of questionnaire interview, education. The NSHD, SALIA and SAPALDIA contributed information from two assessment rounds. Ethical approvals for analyses were obtained for all cohorts.

Outcome definition Chronic bronchitis symptoms were assessed by questionnaire in all cohorts using standard questions based on those defined by the British Medical Research Council (MRC) in 1965,1 as reported cough and phlegm production first thing in the morning and/or during the day or at night for 3 months of the year for ≥2 years. Other outcomes investigated were chronic cough (reported cough for 3 months for ≥2 years regardless of reported phlegm or not) and chronic phlegm (reported phlegm 2

for 3 months for ≥2 years regardless of reported cough or not), except for SALIA, where questions regarding phlegm production were not asked separately from cough, therefore it was not possible to derive the outcome of ‘chronic phlegm’ (online supplements-2).

Exposure measurements ESCAPE-period exposures The ESCAPE exposure assessments have been described elsewhere.28 29 Briefly, a standardised protocol was applied in all geographic sites within the ESCAPE areas during October 2008 to April 2011. Nitrogen dioxide (NO2) and nitrogen oxides (NOx) measurements were conducted in 36 ESCAPE study areas while PM (PM10, PM with aerodynamic diameter ≤10 mm and PM2.5, PM with aerodynamic diameter ≤2.5 mm) were measured in 20 ESCAPE study areas, both in a 14-day period of each of three seasons (cold, warm and intermediate). Annual average concentrations for each monitoring site were calculated by combining the three 14-day periods with measurement data from a centrally located reference site, in operation during the whole study period, to adjust for temporal variability. The land-use regression (LUR) model developed used geographic information system (GIS)-derived predictor variables to describe spatial variation of annual average concentrations for each study area at measurement locations. An annual average estimate was then assigned from the LUR models to each geocoded address (place of residence) based on the date of questionnaire assessment for study participants. In addition, two indicators of local exposures to traffic were derived for each participant’s address: traffic intensity on the nearest road (traffic intensity, vehicles/ day) and total traffic load on major roads in a 100 m buffer (traffic load, vehicles*m/day). Each participant was assigned an annual average concentration at home outdoor of NO2, NOx and the background levels of NO2. For participants residing within ESCAPE areas with PM measurements, they were also assigned exposures to PM2.5, PM10, the coarse fraction of PM (PMcoarse as PM10 minus PM2.5) and PM2.5abs, the light absorbance of PM2.5 (similar to ‘black carbon’).

Back-extrapolation Questionnaire assessments in some cohorts occurred prior to the ESCAPE monitoring campaign in 2008–2011, with some up to 25 years earlier. Due to changes (usually decreases) in air pollution over time, ESCAPE-period exposure values were backextrapolated to the years of collection of health data assuming proportional changes in within-city spatial patterns. Here, individually assigned estimates of ambient concentrations were adjusted (calibrated) for the long-term trends using a predefined backextrapolation algorithm (see http://www.escapeproject.eu/manuals/ Procedure_for_extrapolation_back_in_time.pdf; accessed 10 May 2014). Back-extrapolation for NO2 and PM10 was conducted by ratio methods to the most recent follow-up years in ECRHS and SAPALDIA (assessments in 1998–2002 and 2002, respectively) and also to earlier assessment in SALIA in 1985–1994, NSHD in 1999 and SAPALDIA in 1991.

Statistical analyses Each cohort was first analysed separately using centrally written analytic codes, and harmonised outcome and confounder variables. Descriptive analyses were conducted including Spearman correlation coefficients. The analytic strategy, including all models, sensitivity and subgroup analyses, was specified a priori, Cai Y, et al. Thorax 2014;0:1–10. doi:10.1136/thoraxjnl-2013-204352


Environmental exposure based on current knowledge. Cross-sectional analyses using logistic regression models were undertaken to obtain cohortspecific ORs. Results were then combined using both fixed effects and random effects meta-analyses; pooled estimates from the latter were only shown when heterogeneity ( p